Search

Piles (kinds of geotechnical structures) are used for resisting various lateral loads including earthquakes and inclined loads. Hence, these structures' behavior under lateral load should be studied. Therefore, this investigation studies the lateral deflection (LD) of piles under different situations. 192 physical models were carried out by consideration of the most important factor on the lateral deflection amounts in dried sandy soils. Besides, a model of the Elman Neural Network (ENN) - Improved Arithmetic Optimizer (IAO) algorithm was suggested for predicting the piles' lateral deflection. For the intention of comparison, the Elman Neural Network model and Particle Swarm Optimization - Artificial Neural Network were utilized in lateral deflection amounts estimation. For evaluating the proposed model validity, some parameters like Variance Account For, determination coefficient, and Root Mean Squared Error were estimated. The results showed the ENN-IAO method is more reliable for lateral deflection prediction in a small-scale pile in comparison to the ENN method and PSO-ANN model.

Molecular separation using nanostructured materials have attracted much attention recently for various applications. Different classes of materials have been used among which metal organic framework (MOF) and layered double hydroxide (LDH) have been recently developed due to their superior structure in separation, specifically adsorption process. In this work, we have studied removal of dye from water using a nanocomposite of MOF/LDH. The method of investigation is development of an artificial intelligence-based model for prediction of the adsorption process. The adsorption data have been obtained for removal of a dye (orange II reactive dye) from water at different conditions. The model was proposed using artificial neural network for simulation of the adsorption output which was considered to be equilibrium concentration in the solution (Ce). Indeed, the equilibrium concentration of solute was assumed as the main output in developing the model, while two inputs were postulated including the pH and adsorbent dosage. The model was built considering two hidden layers in the neural network. The validation and training steps were carried out and statistical analysis indicated excellent agreement between the simulated and measured data possessing high coefficient of determination (R2 greater than 0.999). The model revealed to be a high-performance model for simulation of dye removal using adsorption process with high accuracy.